7/12/2006

Computer Science Report 1st Edition

The following is a report I have written for my introductory computer science class that involves the explanation of how to display a scene captured in 3 dimensions to a 2-dimensional environment. I'm not quite sure if how I explained it works out; I may have missed a few fine details here and there, but this is the gist I got from the lecture.

Assignment 5: Multimedia

How might a 3-dimensional scene be reconstructed in a 2-dimensional environment? Well, first we must capture the 3-dimensional data. This can be done by using a controlled environment with multiple cameras at several angles called a motion-capture studio. If we need to, say, capture the motion of a human being walking, we must capture the key points of the human body as it moves. Feet, knees, elbows, arms, legs, head, neck, and other joints must be highlighted somehow so the cameras know to track only those motions.

Once the camera captures these motions in relation to some common reference point, they must be sent to a computer for processing. Now it is time to come up with some way to map the 3-dimensional information into a 2-dimensional form that the computer can process and sort. For each pivot on the joints in which we want to monitor we can map them out in an x, y, z plane in relation to that same reference point we used to capture the data, but that is still in three dimensions. We can, however, say that reference point is the origin on the x, y, z plane and then assign each point of the joints a specific coordinate value throughout the entire time of the capture session. We now have a numerical value in this three dimensional space to work with, which means we can now pinpoint one point on the graph using a set of 2-dimensional numbers – one x, one y, and one z coordinate, but how can this be done for the entire slew of data that we need to display the motions of a human walking?

The answer lies in an operator named the matrix. Matrices can store vast amounts of data in a table format and display it in a 2-dimensional environment. Since now we have three numerical values (the x, y, and z coordinates) of a certain joint we need to map at a specific time in the motion capture data (let’s say the first hundredth of a millisecond), we can place these data inside of a matrix or a series of matrices (one for each individual pivot point and joint). The columns of the matrices can display the x, y, and z coordinates of the points we want to examine, and the rows of the matrices can represent the time in milliseconds (or any time value) of the data from the motion capture process. Using this data from matrices, it can then be reconstructed back into a 3-dimensional model by the computer to display back the main motions of a human walking.

0 Comments:

Post a Comment

<< Home